Professional Documents
Culture Documents
Rule Governed Behavior
Rule Governed Behavior
Rule Governed Behavior
Rule-governed behavior is generally considered an integral component of complex verbal repertoires but
has rarely been the subject of empirical research. In particular, little or no previous research has
attempted to establish rule-governed behavior in individuals who do not already display the repertoire.
This study consists of two experiments that evaluated multiple exemplar training procedures for teaching
a simple component skill, which may be necessary for developing a repertoire of rule-governed behavior.
In both experiments, children with autism were taught to respond to simple rules that specified
antecedents and the behaviors that should occur in their presence. In the first study, participants were
taught to respond to rules containing ‘‘if/then’’ statements, where the antecedent was specified before the
behavior. The second experiment was a replication and extension of the first. It involved a variation on
the manner in which rules were presented. Both experiments eventually demonstrated generalization to
novel rules for all participants; however variations to the standard procedure were required for several
participants. Results suggest that rule-following can be analyzed and taught as generalized operant
behavior and implications for future research are discussed.
Key words: rule-governed behavior, rule-following, instructional control, conditionality, autism,
relational frame theory
Applied behavior analysis is a science that having to directly contact the contingencies,
endeavors to solve problems involving so- that is, without ever having to engage in the
cially important behavior by identifying the behavior of drinking bleach or of contacting
variables of which such behavior is a the consequence of dieing. By definition, one
function, thereby allowing for its prediction can ‘‘follow’’ a rule, without ever having
and control (Cooper, Heron, & Heward, contacted the contingencies that it describes.
2007). Behavior may be easiest to control Skinner (1974) described RGB as partic-
when the environmental variables, of which ularly crucial for the existence and mainte-
it is a function, are readily apparent and/or nance of human civilization. RGB is impor-
are to be found in the recent history of the tant because it allows humans to respond
person (Skinner, 1974). A special class of effectively in life without having to directly
behavior, however, defies efforts at the contact contingencies that would be destruc-
identification of immediately apparent envi- tive or inefficient to contact. Rules allow one
ronmental contingencies that are responsible to avoid dangerous consequences for behav-
for its occurrence, namely, that of rule- ior (e.g., the rule ‘‘Look both ways before
governed behavior (RGB). RGB is behavior crossing the street’’). Rules also allow one to
that occurs due to contact with rules that profit from the experience of previous
describe contingencies, and not due to prior generations by contacting rules that previous
contact with the contingencies the rule generations have derived through their con-
describes (Skinner, 1969). For example, one tact with contingencies. The laws of science
can respond effectively to the rule ‘‘If you are such rules (Skinner, 1969, 1974). For
drink bleach, you will die,’’ without ever example, the principle of reinforcement can
be taught to a university student or clinician
and it can be applied immediately. It is not
The authors would like to thank Ben Craighead necessary for each new person to discover
for his contributions to the study. Address corre-
spondence to Jonathan Tarbox, PhD, BCBA-D, the principle of reinforcement through ran-
19019 Ventura Blvd, 3rd Floor, Tarzana, CA dom contact with contingencies in the
91356. (E-mail: j.tarbox@centerforautism.com). laboratory. The same may be said of the
125
126 JONATHAN TARBOX et al.
laws of physics, biology, chemistry, and research was an important first step in an
engineering. Each new engineer need not empirical examination of RGB but concep-
discover through direct contact with the tual and empirical research had generally not
consequences of their behavior how to build attempted a functional analysis of the
a bridge that will not fall down. RGB is controlling variables involved in RGB, nor
therefore among the most important, foun- an analysis of how the ability to follow rules
dational classes of behavior for human may be established (Hayes, Blackledge, &
civilization and modern life, as we know it Barnes-Holmes, 2001).
would be impossible without it. In an early conceptual treatment of
Despite the importance of RGB, relatively reference and understanding, Parrott (1984)
little attention has been paid to it in the argued that the behavior of simply complying
behavior analytic literature. While Skinner’s with a rule is not equivalent with under-
books include the bulk of early behavioral standing the rule. A person may hear a rule
work on the topic, his conceptual analyses of and not understand it (as in hearing a rule in a
RGB vary somewhat across his writings. In foreign language) or may hear a rule,
some places, Skinner described rules as understand it, and not comply for other
stimuli that altered the operant and/or reasons (e.g., the person has no history of
respondent properties of other stimuli complying with rules stated by that particular
(1957, p. 359), sometimes referring to these speaker, etc.). In the first case, the person
as ‘‘contingency-specifying stimuli’’ (1969, does not understand the rule, in the second
p. 169), although specification, per se, was case, he does. Put another way, in the first
generally left unaddressed (Hayes, 1991; case, the rule specifies nothing, while in the
Parrott, 1984). In other places, Skinner second case, the contingencies specified by
described rules as discriminative stimuli the rule are clear (albeit, not effective).
(1969, p. 148), presumably because they Conceptual work in recent years has
control behavior ‘‘as though’’ they were analyzed RGB from the perspective of
discriminative stimuli, despite the lack of an relational frame theory (RFT). Space does
appropriate history of differential reinforce- not permit a full conceptual treatment of an
ment that defines the concept of the discrim- RFT analysis of RGB (see Barnes-Holmes,
inative stimulus. Regardless, the general O’Hora, Roche, Hayes, Bisset, & Liddy,
thrust of Skinner’s writings suggested that 2001; Tarbox, Tarbox, & O’Hora, 2009), but
people engage in RGB because they have a a brief overview should suffice. The founda-
history of reinforcement for doing so (1957, tion of an RFT analysis of RGB is consistent
1969, 1974). Generally, then, it might be with Skinner’s basic position, namely, that
stated that rule-following can be conceptual- rule-following can be conceptualized as an
ized as a class of behavior, in itself. operant. However, RFT provides further
Empirical research on RGB was active in elaboration on an analysis of the behavior/
the 1980s, centering around the work of environment relations involved in RGB, as
Charles Catania and colleagues. The research well as analyzing how such a repertoire can
conducted in that period generally focused on be acquired, as we describe below.
such topics as identifying whether operant Relational frame theory conceptual treat-
behavior was primarily rule-governed or ments of RGB have provided a functional
contingency-shaped (Shimoff, Mathews, & analysis of what it means when a rule
Catania, 1986), comparing effects of rules ‘‘specifies’’ a contingency. The environmen-
that described performance versus those that tal events, which a rule ‘‘specifies’’ for any
described contingencies (Mathews, Catania, given person, are the environmental variables
& Shimoff, 1985), or studying the differences that participate in relational frames with the
between the properties of rule-governed and stimuli in the rules, for the particular person
contingency shaped behavior, such as differ- listening to the rule. Consider the rule ‘‘If it’s
ences in sensitivity to changes to contingen- raining, then take an umbrella, and you won’t
cies (Shimoff, Catania, & Mathews, 1981; get wet.’’ The stimulus ‘‘raining’’ partici-
see also Hayes, 1989, for a collection of pates in an equivalence relation (or frame of
conceptual papers commenting on early coordination) with the actual sights and
empirical work on RGB). This body of sounds of rain, the stimulus ‘‘take an
RULE-GOVERNED BEHAVIOR 127
preschool children. Thus, the initial evidence program in his/her home. The child was
supporting the use of MET for establishing seated at a table with the behavioral therapist
relational operants is encouraging, and there who was implementing the teaching pro-
is no reason to believe relational operants of gram. An additional behavioral therapist or
conditionality might not also be amenable to research assistant was often present in order
instruction via MET. to collect interobserver agreement data and to
The purpose of the current study was to ensure treatment integrity.
investigate a procedure for establishing a Prior to the study, probes were conducted
generalized ability to respond to simple rules to confirm that the participants could cor-
in children with autism who displayed no rectly tact and receptively identify each of
evidence of a rule-following repertoire. In the stimuli used in the study. Probes were
specific, two experiments investigating var- also conducted to ensure that the participants
iations on an MET procedure were conduct- could correctly respond to the simple in-
ed, in which children with autism were structions used in the study such as, ‘‘Clap
taught to respond to rules specifying ante- your hands.’’ Correct responding to these
cedents and behaviors. The critical outcome probes resulted in verbal praise. One probe of
of the study was to demonstrate generaliza- each tact and each instruction was conducted.
tion of the ability to follow rules for which Stimuli and instructions were only included
participants had never contacted the specified in the study if the participant responded
contingencies, the defining characteristic of correctly to the single probe for it.
RGB.
Response Measurement and
EXPERIMENT 1 Interobserver Agreement
Table 1
Rules Presented During Baseline, Training, and Generalization Probes in Experiment 1
During half of the trials, the stimulus session, according to the following most-to-
described in the rule was presented. During least prompt fading hierarchy: (1) full
the other half of the trials, a stimulus that was physical: the participant was physically
not described in the rule (but which were guided to emit the motor response, (2) partial
described in different rules on other trials) physical: the therapist used light physical
was presented. For example, on one trial, a touch to guide the participant to emit the
picture of a car might be held up and the motor response, (3) model: the therapist
rule presented ‘‘If this is a carrot, then demonstrated the motor response, (4) vocal:
clap,’’ whereas on a later trial, the same the therapist vocally stated the motor re-
stimulus (picture of car) would be presented, sponse, and (5) no prompt. All correct
along with the rule, ‘‘If this is a car, then responses were reinforced, regardless of
wave.’’ In other words, each rule was whether they followed a prompt. Contingent
presented an equal number of times with on an incorrect response, the therapist stated
the stimulus specified in the rule as absent ‘‘no’’ in a neutral tone of voice, and provided
versus present, and each stimulus included descriptive feedback, such as, ‘‘I said if this
in the experiment was presented an equal is a carrot then clap but look that’s not a
number of times with the rule that described carrot so don’t clap.’’ If a participant began
it and the rules that did not describe it. Each to respond before the entire rule was stated,
session consisted of 12 trials, comprised of 6 therapists used partial physical guidance (i.e.,
rules. Each rule was presented for one trial light physical touch to the participant’s
with the antecedent stimulus specified in the hands) so participants would place their
rule present and one trial with the antecedent hands on their lap. Each time a new rule
stimulus absent (6 rules 3 once present plus was introduced, the prompt-fading hierarchy
once absent 5 12 trials). The order of rules was initiated at the highest level of prompt-
was random. Trials of mastered items were ing and prompts were faded within-session.
interspersed and the child received rein- Most-to-least prompt fading was continued
forcement for correct responses to mastered on subsequent sessions, until a participant
items in order to maintain general compli- demonstrated correct independent responding
ance. on two trials: one trial where the specified
Training. During training, a picture card stimulus was present and one where it was
and rule were presented. Correct responses absent. After meeting this criterion, during
were followed by a preferred item selected subsequent sessions where the same rule was
via a brief multiple stimulus preference continuing to be trained, the same prompting
assessment (DeLeon & Iwata, 1996) con- hierarchy was used, but was implemented in
ducted prior to each session. Prompts for reverse order, according to a within-session,
engaging or not engaging in the behaviors least-to-most sequence. Training sessions
specified in the rules were faded out, within- consisted of 10 trials.
130 JONATHAN TARBOX et al.
During half of the training trials the stimulus probes, except that rules were stated with the
described in the rule was presented. In the behavior specified before the antecedent, as
other half of the trials a different stimulus was in the altered rule presentation format phase,
presented. While the first rule was being described above.
trained, sessions consisted of 10 trials. Half
of these trials contained the stimulus described Results
in the rule and the other half contained stimuli
not described in the rule. After the first rule Figure 1 depicts the results of Experiment
was acquired, training on it was terminated 1 and the top panel depicts results for David.
and training was then conducted on the second Note that participants had already learned
rule in the same manner. After the second rule how to perform all the actions described in
was acquired, all subsequent sessions were 12 the rules prior to the start of the experiment,
trials long. Each time an additional new rule so it was expected that participants would
was introduced in training, six trials were exhibit these actions frequently, even in
allocated to it, with three trials allocated to baseline. Therefore, because the stimulus
each of the last two mastered rules. Trials specified by the rule was present during
rotated randomly between the three rules that 50% of trials, if the child simply emitted
comprised the session. whatever motor response was described in
During all training sessions, the criterion the rule on every trial, regardless of whether
for mastery of a particular rule was set at the specified antecedent stimulus was in
80% or more correct across two consecutive place, his behavior would be 50% correct.
sessions. In addition, the participant had to Accordingly, as the top panel of Figure 1
respond correctly the first time the stimulus depicts, David’s correct responding was
in the rule was present and the first time the consistently low in baseline (25–42%). David
stimulus in the rule was not present, during acquired the first rule in the training phase
each session. Once criteria had been met, a after 15 sessions. Generalization was then
generalization probe was conducted. probed and found to be absent. Additional
Generalization probes. Generalization rules were then trained and generalization
probes were identical to baseline and includ- was again probed after each rule was
ed only rules that had not been trained. If the acquired. Generalization was not clearly
participant scored below 80% correct on the demonstrated until David was trained on 11
generalization probe, training continued with rule exemplars. Correct responding main-
the next rule. If the participant scored over tained at the one and two week follow-up
80% on the generalization probe, training generalization probes.
was discontinued and follow-up probes were The middle panel of Figure 1 depicts the
conducted after one and two weeks. The two results for Frank. Frank’s baseline data
week follow-up probe was conducted by a indicate responding between 42 and 58%
different therapist to show generalization correct. Frank met criteria for generalization
across time (maintenance) and generalization after three exemplars were directly trained but
across people. maintenance was not demonstrated at the one
Altered rule presentation format. This week follow-up. A second follow-up probe
phase was identical to the training phase, was then conducted and correct responding
except in how rules were stated. Rules were remained low, so Frank was trained in
presented so that the behavior was described additional exemplars. After Frank was trained
before the antecedent was specified, as in on four additional exemplars and still did not
‘‘Clap if this is a carrot,’’ instead of ‘‘If this is demonstrate generalization, the altered rule
a carrot, then clap.’’ This phase was intro- presentation format phase was initiated. In
duced because it was hypothesized that this addition, after each rule exemplar was mas-
format of rule presentation may make the tered, generalization probes were conducted
antecedent described in the rule more salient according to the altered generalization probe
because it is the last stimulus presented in the format. Frank met criteria for generalization
rule. after two additional exemplars were trained
Altered generalization probe format. This and correct responding remained high at the
phase was identical to the generalization one and two week follow-up probes, therefore
RULE-GOVERNED BEHAVIOR 131
Figure 1. Percentage of correct responses across all conditions of Experiment 1 for David, Frank, and Joey.
Table 2
Rules Presented During Baseline, Training, and Generalization Probes in Experiment 2
Figure 2. Percentage of correct responses across all conditions of Experiment 2 for Jeremy, Tim, and
Greg.
MET intervention. However, this possibility Another potential strategy for avoiding
appears highly unlikely, given the multitude participant variables, which may prevent the
of trials to which they responded incorrectly acquisition of RGB, may be to include
to similar rules specifying similar stimuli typically developing children as participants,
during baseline and throughout intervention. rather than children with developmental
Nevertheless, future studies should include disabilities. Typically developing children
one trial of each rule to be later used in a first do not require explicit intervention in order
trial generalization probe, with no pro- to develop repertoires of RGB, so such
grammed consequence for correct or incor- research may be less socially valid, but it
rect responding, during the baseline phase. may provide a more convenient research
A further potential limitation of this study context in which to study the basic processes
is the fact that experimenters responded by involved in the establishment of rule-follow-
saying ‘‘okay’’ in a neutral tone of voice, ing repertoires.
regardless of participant response during A significant limitation to the current
baseline. It is possible that this consequence study is the fact that, although generalization
was not actually neutral and could have to novel rules was demonstrated for all
served as positive reinforcement during participants, no attempt was made to assess
baseline. However, this possibility seems generalization of rule-following to rules that
particularly unlikely, given that no upward participants contacted in the course of their
trends were observed in accuracy during day-to-day life. That is, generalization of the
baseline. Furthermore, even if this conse- basic ability to understand and respond to if/
quence was a reinforcer, it was delivered then contingency statements was demonstrat-
noncontingently, so it is unlikely that it ed but it is not known if generalization
would have strengthened correct responding occurred on a broader basis. The purpose of
anymore than it would have strengthened the current two experiments was to conduct
incorrect responding. Indeed, in a compari- an initial evaluation of whether establishing a
son of various post-testing procedures in a basic component skill of rule-following was
stimulus equivalence experiment, LeBlanc, possible via MET, not to assess real-life
Miguel, Cummings, Goldsmith, and Carr generalization, however future research
(2003) found that the inclusion of non- should attempt to ensure that treatment gains
contingent reinforcement during post-testing are applied across participants’ everyday
produced similar results as when post-testing lives.
was conducted under extinction. The two experiments in the current study
An additional limitation of the two exper- demonstrated that MET can be used to
iments is that the results obtained were establish the generalized ability to follow
significantly idiosyncratic across partici- simple rules, containing if/then contingency
pants. One participant in each experiment statements that describe antecedents and
readily demonstrated generalization to novel behaviors. This is the first study, of which
rules after being trained on a small number of the authors are aware, where the primary
exemplars. However, the other four partici- purpose of the study was to establish RGB in
pants either required training on many individuals who do not already display it.
exemplars across a long period of time Further, the results of this study demonstrate
(David) or required a modification to the that such a repertoire can be established in
basic procedure (Frank, Tim, and Greg). It is children with autism. However, this study is
not possible to determine the cause of the not without its limitations and future research
idiosyncratic results from the current data but is needed to identify prerequisite skills so
it was likely due to differences among that participants can be appropriately
reinforcement histories and current reper- matched to training procedures and more
toires of the participants at the time the consistent positive results can be obtained.
studies were initiated. For example, there are
likely prerequisite skills that are necessary REFERENCES
before MET in rule-following is likely to be
successful. Future research should attempt to Barnes-Holmes, D., O’Hora, D., Roche, B.,
empirically identify what these skills are. Hayes, S. C., Bisset, R. T., & Liddy, F.
138 JONATHAN TARBOX et al.
(2001). Understanding and verbal regula- Barnes-Holmes, & B. Roche (Eds.), Re-
tion. In S. C. Hayes, D. Barnes-Holmes, & lational frame theory: A post-Skinnerian
B. Roche (Eds.), Relational Frame account of human language and cognition
Theory: A Post-Skinnerian Account of (pp. 21–50). New York, NY: Kluwer.
Human Language and Cognition (pp. Horne, P. J., & Lowe, C. F. (1996). On the
103–118). New York, NY: Kluwer. origins of naming and other symbolic
Barnes-Holmes, Y., Barnes-Holmes, D., behavior. Journal of the Experimental
Roche, B., & Smeets, P. (2001). Exemplar Analysis of Behavior, 65, 185–241.
training and a derived transformation of Kazdin, A. E. (1982). Single-case research
function in accordance with symmetry: II. designs. New York: Oxford University
The Psychological Record, 51, 589–604. Press.
Cooper, J. O., Heron, T. E., B Heward, W. L. LeBlanc, L. A., Miguel, C. F., Cummings, A.
(2007). Applied behavior analysis. (2nd R., Goldsmith, T. R., & Carr, J. E. (2003).
ed.). Upper Saddle River, NJ: Pearson The effects of three stimulus-equivalence
Education. testing conditions on emergent US geog-
DeLeon, I. G., & Iwata, B. A. (1996). raphy relations of children diagnosed with
Evaluation of a multiple-stimulus format autism. Behavioral Interventions, 18,
for assessing reinforcer preferences. Jour- 279–289.
nal of Applied Behavior Analysis, 29, Mathews, B. A., Catania, A. C., & Shimoff,
519–533. E. (1985). Effects of uninstructed verbal
Fiorile, C. A., & Greer, R. D. (2007). The behavior on nonverbal responding: Con-
induction of naming in children with no tingency descriptions versus performance
prior tact responses as a function of descriptions. Journal of the Experimental
multiple exemplar histories of instruction. Analysis of Behavior, 43, 155–164.
The Analysis of Verbal Behavior, 23, 71– O’Hora, D., Barnes-Holmes, D., Roche, B.,
87.
& Smeets, P. (2004). Derived relational
Greer, R. D., Stolfi, L., Chavez-Brown, M., networks and control by novel instruc-
& Rivera-Valdez, C. (2005). The emer-
tions: A possible model of generative
gence of the listener to speaker component
verbal responding. The Psychological
of naming in children as a function of
Record, 54, 437–460.
multiple exemplar instruction. The Anal-
Parrott, L. J. (1984). Listening and under-
ysis of Verbal Behavior, 21, 123–134.
standing. Behavior Analyst, 7, 29–39.
Hayes, L. J. (1991). Substitution and refer-
ence. In L. J. Hayes & P. N. Chase (Eds.), Shimoff, E., Catania, A. C., & Mathews, B.
Dialogues on verbal behavior: The First A. (1981). Uninstructed human respond-
International Institute on Verbal Relations ing: Sensitivity of low-rate performance to
(pp. 3–14). Reno, NV: Context Press. schedule contingencies. Journal of the
Hayes, S. C. (Ed.). (1989). Rule-governed Experimental Analysis of Behavior, 36,
behavior: Cognition, contingencies, and 207–220.
instructional control. New York, NY: Shimoff, E., Mathews, B. A., & Catania, A.
Plenum Press. C. (1986). Human operant performance:
Hayes, S. C., Blackledge, J. T., & Barnes- Sensitivity and pseudosensitivity to con-
Holmes, D. (2001). Language and cogni- tingencies. Journal of the Experimental
tion: Constructing an alternative approach Analysis of Behavior, 46, 149–157.
within the behavioral tradition. In S. C. Skinner, B. F. (1957). Verbal behavior.
Hayes, D. Barnes-Holmes, & B. Roche Acton, MA: Copley Publishing Group
(Eds.), Relational frame theory: A post- and the B. F. Skinner Foundation.
Skinnerian account of human language Skinner, B. F. (1969). Contingencies of
and cognition (pp. 3–20). New York, NY: reinforcement: A theoretical analysis.
Kluwer. New York: Appleton-Century-Crofts.
Hayes, S. C., Fox, E., Gifford, E. V., Wilson, Skinner, B. F. (1974). About behaviorism.
K. G., Barnes-Holmes, D., & Healy, O. New York, NY: Vintage Books.
(2001). Derived relational responding as Stokes, T. F., & Baer, D. M. (1977). An
learned behavior. In S. C. Hayes, D. implicit technology of generalization.
RULE-GOVERNED BEHAVIOR 139